21793 research outputs found
Sort by
Detection of COVID-19 using Deep Learning Techniques and Extraction of the Infected Region using Lung Image Segmentation
Get PDFThe world at present is still suffering from COVID-19 pandemic. COVID-19 is caused by SARS-CoV-2, which damages human lungs and results in pneumonia. Pneumonia is a disease in which the lungs become filled with fluid and inflame, leading to breathing difficulties. Sometimes, breathing problems become very severe, requiring proper treatment, including oxygen or a ventilator.
Though RT-PCR tests are the commonly used method to detect COVID-19 virus, radiological tests are often used by doctors to diagnose the disease based on severity level and risk factors. This thesis concentrates on two major issues, automatic detection of the COVID-19 infection using deep learning techniques and the determination of the severity level of the infection to help reduce the manual tasks and burden of the radiologist.
In the first part, different deep neural network architectures including convolutional neural network-based ResNet50, DarkNet19, GoogLeNet, and VGG16 methods along with a self-attention based vision transformer (ViT) approach called COViT-CT are implemented to detect COVID-19 CT-Scan images. The performance of the various architectures are compared using various metrics, such as accuracy, precision, recall (sensitivity), specificity, F1-score and AUC, as well the confusion matrix, and the best architecture with the highest accuracy, which is COViT-CT, is selected for automatic COVID-19 detection.
In the second part, if the CT-Scan image is COVID-19 positive during the first part, then the image segmentation method is used to extract the COVID-19 infected region from the Lung CT-Scan images. The infected region is useful in determining the severity level of the COVID-19 infection, which helps in the diagnosis of the disease.
All the experiments are performed using the SARS-CoV-2 CT-Scan dataset. It is shown that the self-attention based COViT-CT method provides the best performance on the test sets of the above-mentioned dataset
Graph Neural Networks For 3D Human Pose Estimation
Get PDFIn human pose estimation methods based on graph convolutional architectures, the human skeleton is usually modeled as a graph whose nodes are body joints and edges are connections between neighboring joints. However, most of these methods tend to focus on learning relationships between body joints of the skeleton using first-order neighbors, ignoring higher-order neighbors and hence limiting their ability to exploit relationships between distant joints. In this thesis, we introduce a higher-order regular splitting graph network (RS-Net) for 2D-to-3D human pose estimation using matrix splitting in conjunction with weight and adjacency modulation. The core idea is to capture long-range dependencies between body joints using multi-hop neighborhoods and also to learn different modulation vectors for different body joints as well as a modulation matrix added to the adjacency matrix associated to the skeleton. This learnable modulation matrix helps adjust the graph structure by adding extra graph edges in an effort to learn additional connections between body joints. Instead of using a shared weight matrix for all neighboring body joints, the proposed RS-Net model applies weight unsharing before aggregating the feature vectors associated to the joints in order to capture the different relations between them. Experiments and ablations studies performed on two benchmark datasets demonstrate the effectiveness of our model, achieving superior performance over strong baselines for 3D human pose estimation.
The other contribution of this thesis consists of designing a spatio-temporal 3D human pose estimation model using multilayer perceptrons and graph neural networks. Despite the success of graph convolutional networks and their variants in 3D human pose estimation tasks, most of these methods only consider spatial correlations between body joints and do not take into account temporal correlations, thereby limiting their ability to capture relationships in the presence of occlusions and inherent ambiguity. To address this issue, we propose a spatio-temporal network architecture composed of a joints-mixing multi-layer perceptron block that facilitates communication among different joints and a graph weighted Jacobi network block that enables communication among
various feature channels. Extensive experiments on two benchmark datasets demonstrate the competitive performance of our model, outperforming recent state-of-the-art methods for 3D human pose estimation. In addition, we perform a runtime analysis and conduct a comprehensive ablation study to show the effect of the key components of our model
Diversity Awareness in Software Engineering Participant Research
Get PDFDiversity and inclusion are necessary prerequisites for shaping technological innovation that benefits society as a whole. A common indicator of diversity consideration is the representation of different social groups among software engineering (SE) researchers, developers, and students. However, this does not necessarily entail that diversity is considered in the SE research itself.
Our study examines how diversity is embedded in SE research, particularly research that involves participant studies. To this end, we selected 79 research papers containing 105 participant studies spanning three years of a renowned SE conference. Using a content analytical approach, we identified how SE researchers report the various diversity categories of their study participants and investigated: 1) the extent to which participants are described, 2) what diversity categories are commonly reported, and 3) the function diversity serves in the SE studies.
Our results demonstrate that even though most SE studies report on the diversity of participants, SE research often emphasizes professional diversity, such as occupation and work experience, over social diversity, such as gender or location of the participants. Furthermore, our results show that participant diversity is seldom analysed or reflected upon when SE researchers discuss their study results, outcomes, or limitations.
To help researchers self-assess their study diversity awareness, we propose a diversity awareness model and guidelines that SE researchers can apply to their research. Furthermore, our research contributes to the United Nations Sustainable Development’s Goals 5, 9, and 10. With this study, we hope to shed light on a new approach to tackling the diversity and inclusion crisis in the SE field
The CAQ and immigration: a new frontier for Quebec politics?
Get PDFIn 2018, when the CAQ was elected to form the new government of Quebec, it won on a platform that contained numerous measures to restrict immigration. Some of these measures were criticized and described as marking a radical shift in the province’s historical approach to immigration. In order to gain a more precise understanding of the implications of the CAQ’s position for Quebec politics, this thesis asks the question: How different is the CAQ’s position on immigration from that of the province’s main other political parties? To answer this question, this thesis looks at electoral platforms and parliamentary debates, using manual coding (NVivo) and computer-aided dictionary analysis (RStudio). By looking at the CAQ, the PLQ and the PQ’s stance, salience and discourse on immigration, it finds that although the CAQ proposed measures that are more restrictive towards immigration, it did so by mobilizing long-standing and well-established discursive logics. This in turn leads us to question our understanding of Quebec as a “pro-immigration” space, as well as the relevance of “pro” and “anti” immigration labels, and invites further research into a more systemized and helpful classification of parties and their positions on immigration
Urban microclimate and its impact on built environment – A review
Get PDFIncreasing urbanization and population growth have brought attention to urban microclimates in recent years. The study on urban microclimate and its impact on the built environment is gaining momentum. A growing number of researchers have examined the relationship between human activity and the immediate surroundings to reduce adverse impacts on the environment and climate. This paper presents the latest progress in urban microclimate research on urban wind and thermal environment, covering traditional methods, including field measurements, wind tunnel modeling, and CFD simulations, as well as emerging methods, such as artificial intelligence or data-driven models. Among the publications reviewed, the topics include isothermal scenarios that neglected thermal aspects (e.g., urban wind energy, wind comfort) and thermal scenarios (e.g., urban heat islands and outdoor thermal comfort). In the review, it was found that CFD has been widely applied due to its well-developed nature. In addition to field measurements, new techniques (such as satellite and thermal imaging) provide valuable validation data for CFD and training data for artificial intelligence applications. In isothermal scenarios, wind tunnel modeling has been successfully applied. However, thermal scenarios present significant challenges. In addition, urban data-driven models have emerged with promising results, but systematic investigations have been limited. In this paper, we identify future research needs for urban microclimates based on an overview of recent progress
Investigating hybrid methods and transfer learning for accurate load forecasting
No full textLoad forecasting with low prediction error is essential to keep minimizing costs in generating and supplying power. It has many applications in energy production, distribution, and infrastructure construction. Because of the high autocorrelation and strong seasonality in load data, it is difficult to build robust and generalizable forecasting models. To address the problem, we propose a hybrid model, the Fourier Split NET (FSNET). The proposed model consists of two phases. A deseasonalization phase where the model uses the Fourier transform to isolate the seasonal component from the data using the fast Fourier transform. The second phase consists of training a simple linear model to replicate the seasonal behavior of the data and training a group of LSTM neural networks on different clusters of the data. The model uses statistical features to build separate LSTM models for different groups of data. We experimented on open datasets and obtained higher accuracy results compared to other forecasting approaches using different accuracy metrics.
In a second contribution, we propose a novel approach for load forecasting that leverages the task affinity score to measure the distance between different tasks. The task affinity score provides a more effective method for measuring the similarity between tasks in a transfer learning context. We demonstrate the efficacy of the task affinity score through empirical analysis using a synthetic dataset. Our results show that the task affinity score outperforms other intuitive metrics such as the loss function for task selection.
To apply this approach, we present the Affinity-Driven Transfer Learning (ADTL) algorithm for load forecasting. The ADTL algorithm optimizes the transfer learning process by leveraging knowledge from pre-trained models and datasets to improve the accuracy of load forecasting for new and previously unseen datasets. We validate the effectiveness of the ADTL algorithm by testing it on two real-world datasets: the Australian Energy Market Operator (AEMO) dataset and the Smart Australian dataset.
Overall, our study highlights the importance of the task affinity score in transfer learning for load forecasting applications. The proposed ADTL algorithm provides a practical solution for improving the efficiency and convergence speed of load forecasting in the energy industry
Star Scientists’ Prediction in the Field of Artificial Intelligence Using Machine Learning Techniques
Get PDFStar scientists are highly influential researchers who have made significant contributions to their field, gained widespread recognition, and often attracted substantial research funding. They are critical for the advancement of science and innovation, and they have a significant influence on the transfer of knowledge and technology to industry. Identifying potential star scientists before their performance becomes outstanding is important for recruitment, collaboration, networking, or research funding decisions. The objectives of this study are to develop a prediction method for star scientists in the artificial intelligence scientific ecosystem and to investigate the features related to their success. Bibliographic data was extracted from Scopus and data mining techniques were employed to gain insights into the authors’ discipline, gender, and ethnicity. The h-index was used as a proxy for research performance, and a dynamic profile of authors was established. Rising stars were found to have different patterns compared to their non-rising stars counterparts in almost all the early-career features. Social network analysis showed that certain features such as gender and ethnic diversity play important role in scientific collaboration and that they can significantly impact an author's career development and success. The prediction of rising stars was based on the author's early-career characteristics such as quantity and quality of research output, metrics obtained from social network analysis, and various diversity measures. Several classifiers in machine learning were trained, tested, implemented, and compared in the prediction task. It was shown that the Random Forest outperformed other classifiers and that the most important combination of features in predicting star scientists in the artificial intelligence field is the number of articles, group discipline diversity, and weighted degree centrality. Our findings highlight the importance of considering the authors' characteristics from different categories of features in the early stages of their careers to identify rising stars. This study offers valuable insights for researchers, practitioners, and funding agencies interested in identifying and supporting talented researchers
Distinctions of Difference: Race, Queerness, & Homonationalism in Québec Cinema
Get PDFDistinctions of Difference: Race, Queerness, & Homonationalism in Québec Cinema
Gregory Pablo Rodríguez-Arbolay, PhD
Concordia University, 2023
This dissertation examines the distinct intertwining of race, queerness, and homo/nationalism in Québec cinema from 1960 to 2003. It traces important historical shifts in Québec and its cinema industries to frame the symbolic and material conditions of sexual and racial representation. Through the conceptual frameworks of queer of colour critique (Muñoz 1999) and homonationalism (Puar 2007), this dissertation questions how cinema has been used as a homo/nationalist tool to promote the notion of a modern and progressive Québec. It argues that Québec’s narrative of LGBTQ+ inclusion regulates visibility and representation in a manner that obscures queers of colour from its cultural imagination.
Chapter 1 examines how the legacies of slavery and colonialism map upon the cultural homophobia of the emergent Québec nationalist project in Claude Jutra’s landmark film, À tout prendre (1963). Chapter 2 traces how the neoliberal collapse of Québec’s nationalist project gave way to new multicultural and sexual politics in Denys Arcand’s celebrated films, Le déclin de l’empire américain (1986) and Les Invasions barbares (2003). Chapter 3 examines the convergence of sexual and multicultural pluralism in Hunt Hoe’s independent film, Seducing Maarya (2000), the first Québec film to spotlight a queer character of colour.
These films collectively chart the historical conditions of homonationalism by elucidating significant shifts in the cultural representation of queer(ed) and racialized subjects in late-twentieth-century Québécois cinema. Through illustrating the rise of LGBTQ+ politics in Québec, I frame how sexual nationalism came to converge with cultural imperialism in Québec cinema; and how it centres white homonormative subjects as symbols of Québec’s modernity. The dissertation contends that this progressive framework limits the possibilities for the representation of diverse racialized sexualities. Through films and historical materials, I challenge how Québécois queer people of colour have been rendered largely absent, silent, or tokenized within our cultural imagination. Overall, the dissertation challenges the notion of visibility and representation as a universal strategy for queers of colour to claim political agency and cultural citizenship
“An Occ-Asian to Make Myself Perfectly Queer”: Exploration of a Queer Asian Music Therapist’s Identities
Get PDFMinoritized music therapists are increasingly represented in music therapy research, but the literature base that includes the experience and expertise of racialized music therapists who also identify as Queer remains small. The purpose of this heuristic self-inquiry was to examine how the researcher’s intersecting Queer and Asian identities and their experiences growing up in western culture inform their early career music therapist identity. Moustakas’ (1990) six phases of heuristic self-inquiry were used to generate and analyze data and to support the final creative synthesis. Qualitative data was generated through journaling and songwriting (with a specific focus on lyric writing) after daily music improvisation sessions that took place over a four week period. Thematic analysis of the data generated three thematic categories with three to four sub-themes each. Categories include: Feelings Associated with Self & Identity Exploration, Challenging Life Experiences, and Generating Momentum. A visual art piece was created to summarize and synthesize the researcher’s experience of the enquiry process. Personal and professional implications were explicated and include a focus on the importance of critical self-reflection, understanding how experiences of marginalization inform our music therapy work, and increasing the representation of minoritized voices in research, education, and practice settings
Thank You for Donating! How Post-Donation Thank-You Notes Impact Charitable Giving
Get PDFCharities depend on individuals’ donations as a main source of revenue. Although there is extensive research on one-time donations, there is a lack of research on recurring charitable giving. Since retaining donors is more cost-effective than acquiring new ones, charities would benefit from better understanding tactics that help increase their donors’ lifetime value. One such tactic is to express gratitude, by sending thank-you notes to donors. However, the effects of thank-you notes on charitable giving are under-researched. The relationship marketing and the donor retention literature both suggest that donors should respond favourably to receiving a thank-you note, but related empirical evidence has provided mixed results. I investigated the effects of thank-you notes on subsequent donations, including their message content (i.e., includes the direct impact on beneficiaries vs. not) and the medium through which they were sent. I conducted three online experiments where I compared the effects of two types of thank-you notes (i.e., basic vs. detailed messaging) to that of a neutral stimulus (i.e., tax receipt), and of two formats (i.e., letter vs. email). I found that the effects of thank-you notes are complex, as there were no direct effects of thank-you notes on attitudes towards the charity and repeat donation amount, but there were indirect effects through two mediators (i.e., perceived effectiveness of the donation and positive emotions towards donating again). A thank-you note’s medium also moderated their effects and had a more consistent impact on attitudes than on donation intentions